A survey: deriving private information from perturbed data

Okkalioglu B. D., Okkalioglu M., KOÇ M., Polat H.

ARTIFICIAL INTELLIGENCE REVIEW, vol.44, no.4, pp.547-569, 2015 (SCI-Expanded) identifier identifier

  • Publication Type: Article / Article
  • Volume: 44 Issue: 4
  • Publication Date: 2015
  • Doi Number: 10.1007/s10462-015-9439-5
  • Journal Indexes: Science Citation Index Expanded (SCI-EXPANDED), Scopus
  • Page Numbers: pp.547-569
  • Keywords: Data reconstruction, Data perturbation, Privacy, Attack resilience, Spectral filtering, RANDOMIZED-RESPONSE, RECONSTRUCTION, ERROR, NOISE
  • Anadolu University Affiliated: Yes


Privacy-preserving data mining has attracted the attention of a large number of researchers. Many data perturbation methods have been proposed to ensure individual privacy. Such methods seem to be successful in providing privacy and accuracy. On one hand, different methods are utilized to preserve privacy. On the other hand, various data reconstruction approaches have been proposed to derive private information from perturbed data. Thus, many researchers have been conducting various studies about data reconstruction methods and the resilience of data perturbation schemes. In this survey, we focus on data reconstruction methods due to their importance in privacy-preserving data mining. We provide a detailed review of the data reconstruction methods and the data perturbation schemes attacked by different data reconstruction techniques. We merge our review with the evaluation metrics and the data sets used in current attack techniques. Finally, we pose some open questions to provide a better understanding of these approaches and to guide future study.